gitbook-docs/tutorials_tutorials/add-your-own-builtin.md
2021-06-18 22:21:42 +00:00

127 lines
3.8 KiB
Markdown

# Add Your Own Builtins
As discussed in the [Node](../knowledge_knowledge/node/knowledge_node_services.md) section, some service functionalities useful to a large audience and such services and can be directly deployed to a peer node. The remainder of this tutorial guides you through the steps necessary to create and deploy a Builtin service.
In order to have a service available out-of-the-box with the necessary startup and scheduling scripts, we can take advantage of the Fluence [deployer feature](https://github.com/fluencelabs/fluence/tree/master/deploy) for Node native Services. This feature handles the complete deployment process including
* module uploads,
* service deployment,
* script initialization and scheduling
Note that the deployment process is a fully automated workflow requiring you to merely submit your service assets in the appropriate structure as a PR to the appropriate GitHub repository. At this point you should have a solid grasp of creating service modules and their associated configuration files. See the [Developing Modules And Services](../development_development/) section for details.
Our first step is fork the ??? repo by clicking on the Fork button, upper right of the repo webpage, and follow the instructions to create a local copy. In your local repo copy, checkout a new branch with a new, unique branch name:
```text
git checkout -b MyBranchName
```
In your new branch create a new directory with the service name in the _builtin_ directory:
```text
cd builtins
mkdir my-new-super-service
cd new-super-service
```
Replace my-_new-super-service_ with your service name.
Now we can build and populate the required directory structure with your service assets. You should put your service files in the corresponding my-_new-super-service_ directory specified in config as `builtins_base_dir`
**TODO: check if that applies to new repo approach.**
#### Requirements
In order to deploy a builtin service, you need
* the Wasm files for each module comprising the service
* a blueprint file specifying service dependencies
* service start or schedule scripts
Blueprints
Just to recap, blueprints specify the service name and module dependencies:
```javascript
// blueprint.json
{
"name": "my-new-super-service",
"dependencies": [
"name: module_1",
"name: module_2"
]
}
```
where _module\_i_ is the name of the Wasm module. Note that dependencies can be specified as string literals or hashes:
```javascript
// blueprint.json
{
"name": "aqua-dht",
"dependencies": [
"hash:558a483b1c141b66765947cf6a674abe5af2bb5b86244dfca41e5f5eb2a86e9e",
"name:aqua-dht"
]
}
```
So modules and configs names should look like this:
```text
-- aqua-dht.wasm
-- aqua-dht_config.json
-- 558a483b1c141b66765947cf6a674abe5af2bb5b86244dfca41e5f5eb2a86e9e.wasm
-- 558a483b1c141b66765947cf6a674abe5af2bb5b86244dfca41e5f5eb2a86e9e_config.json
```
**Start Script**
Start scripts, which are optional, execute once after service deployment or node restarts and are submitted as _air_ files and may be accompanied by a json file containing the necessary parameters.
```text
;; on_start.air
(seq
(call relay ("some_service_alias" "some_func1") [variable1] result)
(call relay ("some_service_alias" "some_func2") [variable2 result])
)
```
and the associated data file:
```javascript
// on_start.json
{
"variable1" : "some_string",
"variable2" : 5,
}
```
**Scheduling Script**
#### Directory Structure
```text
-- builtins
-- {service_alias}
-- scheduled
-- {script_name}_{interval_in_seconds}.air [optional]
-- blueprint.json
-- on_start.air [optional]
-- on_start.json [optional]
-- {module1_name}.wasm
-- {module1_name}_config.json
-- {module2_name}.wasm
-- {module2_name}_config.json
...
```